Lightbulb glowing in dark

Real-time automation identifies high-complexity patients and transforms integrated  care

Machine learning better predicts members at risk of excess forward spend, improving care and case management.

Project Summary

Client: One of the largest integrated health care delivery and financing systems in America.

Challenge: Identify highest complexity patients early and accurately.

Solution: An algorithm that identifies priority patients for integrated treatment plans.

Results: Identified 46,000 high-complexity patients to improve health outcomes and save $56 million in the first year of use.


Receiving hospital treatment is a complicated experience. Throughout their hospital stay, patients engage with scores of administrators, medical specialists, and support staff as the hospital restores them to optimal health. Simply navigating a hospital’s intricate ecosystem can be exhausting and confusing for patients.

The client’s Integrated Care Team makes the treatment journey easier.

Starting on day one, the Care Team provides proactive care and case management for high-complexity patients. Working intimately with patients and hospital staff to design and implement holistic treatment plans, the team helps members address complex health care needs with the best possible outcomes at the lowest possible cost. To provide even more effective service, the Care Team wanted to identify the highest-complexity patients (HCPs) with more speed and accuracy – the faster the team identified and prioritized HCPs, the more impact the care team could have on patient outcomes and cost reduction. The Care Team needed a way to quickly, accurately, and systematically forecast excess forward spend and prioritize highest-complexity patient cases.


To help the Care Team expedite their prioritization process, Lumevity worked with the team to create a Triggering and Routing Machine Learning Model. The machine learning model synthesizes patient data from multiple sources to generate a predictive excess forward spend score for every high-complexity patient in real time.

What makes the machine learning model better than the Care Team’s previous method of prioritization? Simply put, the machine learning model is more efficient, more effective, and more discerning.

From initial receipt of patient authorization to final output of forward spend score, the machine learning model takes mere seconds to do its job. And that’s not for lack of detail or precision – using a method called ensemble learning, the algorithm averages multiple models into a single report. By utilizing the “wisdom of the crowd,” the machine learning model generates the most accurate and comprehensive forecast of excess spend possible.

The machine learning model also uses classification and regression to sift through its massive data collection and pinpoint critical information. Grouping individual data points by shared characteristics and prioritizing the most important of those characteristics, this artificially intelligent model generates spend scores that reflect each patient’s unique medical circumstances.


This innovative collaboration between Lumevity and the Care Team is an inspiring example of what can happen when companies embrace the “and.” At first glance, driving down costs and improving the customer experience seem like fundamentally opposite endeavors. However, by forecasting potential excess spend at an advanced rate, the machine learning model system gave the Care Team the awareness and agility to design care with maximum technological efficiency and deeply human empathy – at the same time.  

Thanks to the machine learning model’s precise scoring and ranking system, the Care Team was able to prioritize the highest-complexity cases first, proactively engage with hospital personnel, and efficiently provide the best possible outcomes at the lowest possible cost.

2x cases identified

The machine learning model identified 88,000 high-complexity cases(compared to 43,000 in the previous year). The automated algorithm more than doubled the Care Team’s diagnostic capacity – in just one year of operation.

$56M savings

In total, the newly streamlined system saved the company $56 million in its first year of use.

You may also like

Analytics governance model liberates resources and activates cross-department engagement

Streamlining analytic procedures and facilitating advanced analytics with organizational/operational modeling

Read More

How to put innovation to work for you

An inside look at a program that’s on track to save $1B in five...

Read More

Team-built automation reduces hours to seconds

Robotic desktop automation creates folders required for biweekly payroll processing

Read More